Fix bbmain to correctly handle #! files

.circleci/config.yml

@@ -55,6 +55,7 @@ templates:
             (cd test/requestconflict/mod6 && go build ./...)
             (cd test/nested && go build ./...)
             (cd test/nested/nestedmod && go build ./...)
+            (cd test/invoke/mod && go build ./...)
       - run:
           name: tests
           command: ./gobuilds.sh
@@ -305,6 +306,7 @@ jobs:
             (cd test/requestconflict/mod6 && go mod tidy && go mod verify)
             (cd test/nested && go mod tidy && go mod verify)
             (cd test/nested/nestedmod && go mod tidy && go mod verify)
+            (cd test/invoke/mod && go mod tidy && go mod verify)
             git status
             if [[ -n "$(git status --porcelain .)" ]]; then
               echo 'go.mod/go.sum is out-of-date: run `go mod tidy` in the right module directories (see git status) and then check in the changes'
@@ -328,6 +330,7 @@ jobs:
             (cd test/requestconflict/mod6 && go vet ./...)
             (cd test/nested && go vet ./...)
             (cd test/nested/nestedmod && go vet ./...)
+            (cd test/invoke/mod && go vet ./...)
       - run:
           name: gofmt
           command: |

gobuilds.sh

@@ -3,6 +3,8 @@ set -eux
 
 # This one hasn't been migrated to the go test yet.
 (cd ./test/requestconflict && ./test.sh)
-
-(cd ./src/cmd/makebb && GO111MODULE=on go build .)
-(cd ./test && go test --makebb=../src/cmd/makebb/makebb -v)
+(cd ./test/nested && ./test.sh)
+(cd ./test/implicitimport && ./test.sh)
+(cd ./test/nameconflict && ./test.sh)
+(cd ./test/12-fancy-cmd && ./test.sh)
+(cd ./test/invoke && ./test.sh)

src/pkg/bb/bbmain/cmd/main.go

@@ -96,3 +96,77 @@ func main() {
 
 	run()
 }
+
+// A gobusybox has 3 possible ways of invocation:
+//
+// ## Direct
+//
+//   ./bb ls -l
+//
+// For the gobusybox, argv in this case is ["./bb", "ls", "-l"] on all OS.
+//
+//
+// ## Symlink
+//
+//   ln -s bb ls
+//   ./ls
+//
+// For the gobusybox, argv in this case is ["./ls"] on all OS.
+//
+//
+// ## Interpreted
+//
+// This way exists because Plan 9 does not have symlinks. Some Linux file
+// system such as VFAT also do not support symlinks.
+//
+//   echo "#!/bin/bb #!gobb!#" >> /tmp/ls
+//   /tmp/ls
+//
+// For the gobusybox, argv depends on the OS:
+//
+// Plan 9: ["ls", "#!gobb!#", "/tmp/ls"]
+// Linux/Unix: ["/bin/bb", "#!gobb!#", "/tmp/ls"]
+//
+// Unix and Plan 9 evaluate arguments in a #! file differently, and, further,
+// invoke the arguments in a different way.
+//
+// (1) The absolute path for /bin/bb is required, else Linux will throw an
+//     error as bb is not in the list of allowed interpreters.
+//
+// (2) On Plan 9, the arguments following the interpreter are tokenized (split
+//     on space) and on Linux, they are not. That means we should restrict
+//     ourselves to only ever using one argument in the she-bang line (#!).
+//
+// (3) Which gobusybox tool to use is always in argv[2].
+//
+// (4) Because of the differences in how arguments are presented to #! on
+//     different kernels, there should be a reasonably unique magic value so
+//     that bb can have confidence that it is running as an interpreter, rather
+//     than on the command-line in direct mode.
+//
+// The code needs to change the arguments to look like an exec: ["/tmp/ls", ...]
+//
+// In each case, the second arg must be "#!gobb!#", which is extremely
+// unlikely to appear in any other context (save testing files, of course).
+//
+// The result is that the kernel, given a path to a #!gobb#! file, will
+// read that file, then exec bin with the argument from argv[2] and any
+// additional arguments from the exec.
+func init() {
+	// Interpreted mode: If this has been run from a #!gobb!# file, it
+	// will have at least 3 args, and os.Args needs to be reconstructed.
+	if len(os.Args) > 2 && os.Args[1] == "#!gobb!#" {
+		os.Args = os.Args[2:]
+	}
+
+	m := func() {
+		if len(os.Args) == 1 {
+			log.Fatalf("Invalid busybox command: %q", os.Args)
+		}
+		// Use argv[1] as the name.
+		os.Args = os.Args[1:]
+		run()
+	}
+	bbmain.Register("bbdiagnose", bbmain.Noop, bbmain.ListCmds)
+	bbmain.RegisterDefault(bbmain.Noop, m)
+}

src/pkg/bb/bbmain_src.go

@@ -1,3 +1,3 @@
 package bb
 
-var bbMainSource = []byte("// Copyright 2018 the u-root Authors. All rights reserved\n// Use of this source code is governed by a BSD-style\n// license that can be found in the LICENSE file.\n\n// Package main is the busybox main.go template.\npackage main\n\nimport (\n\t\"errors\"\n\t\"log\"\n\t\"os\"\n\t\"path/filepath\"\n\n\t\"github.com/u-root/gobusybox/src/pkg/bb/bbmain\"\n\t// There MUST NOT be any other dependencies here.\n\t//\n\t// It is preferred to copy minimal code necessary into this file, as\n\t// dependency management for this main file is... hard.\n)\n\n// AbsSymlink returns an absolute path for the link from a file to a target.\nfunc AbsSymlink(originalFile, target string) string {\n\tif !filepath.IsAbs(originalFile) {\n\t\tvar err error\n\t\toriginalFile, err = filepath.Abs(originalFile)\n\t\tif err != nil {\n\t\t\t// This should not happen on Unix systems, or you're\n\t\t\t// already royally screwed.\n\t\t\tlog.Fatalf(\"could not determine absolute path for %v: %v\", originalFile, err)\n\t\t}\n\t}\n\t// Relative symlinks are resolved relative to the original file's\n\t// parent directory.\n\t//\n\t// E.g. /bin/defaultsh -> ../bbin/elvish\n\tif !filepath.IsAbs(target) {\n\t\treturn filepath.Join(filepath.Dir(originalFile), target)\n\t}\n\treturn target\n}\n\n// IsTargetSymlink returns true if a target of a symlink is also a symlink.\nfunc IsTargetSymlink(originalFile, target string) bool {\n\ts, err := os.Lstat(AbsSymlink(originalFile, target))\n\tif err != nil {\n\t\treturn false\n\t}\n\treturn (s.Mode() & os.ModeSymlink) == os.ModeSymlink\n}\n\n// ResolveUntilLastSymlink resolves until the last symlink.\n//\n// This is needed when we have a chain of symlinks and want the last\n// symlink, not the file pointed to (which is why we don't use\n// filepath.EvalSymlinks)\n//\n// I.e.\n//\n// /foo/bar -> ../baz/foo\n// /baz/foo -> bla\n//\n// ResolveUntilLastSymlink(/foo/bar) returns /baz/foo.\nfunc ResolveUntilLastSymlink(p string) string {\n\tfor target, err := os.Readlink(p); err == nil && IsTargetSymlink(p, target); target, err = os.Readlink(p) {\n\t\tp = AbsSymlink(p, target)\n\t}\n\treturn p\n}\n\nfunc run() {\n\tname := filepath.Base(os.Args[0])\n\terr := bbmain.Run(name)\n\tif errors.Is(err, bbmain.ErrNotRegistered) {\n\t\tif len(os.Args) > 1 {\n\t\t\tos.Args = os.Args[1:]\n\t\t\terr = bbmain.Run(filepath.Base(os.Args[0]))\n\t\t}\n\t}\n\tif errors.Is(err, bbmain.ErrNotRegistered) {\n\t\tlog.SetFlags(0)\n\t\tlog.Printf(\"Failed to run command: %v\", err)\n\n\t\tlog.Printf(\"Supported commands are:\")\n\t\tfor _, cmd := range bbmain.ListCmds() {\n\t\t\tlog.Printf(\" - %s\", cmd)\n\t\t}\n\t\tos.Exit(1)\n\t} else if err != nil {\n\t\tlog.SetFlags(0)\n\t\tlog.Fatalf(\"Failed to run command: %v\", err)\n\t}\n}\n\nfunc main() {\n\tos.Args[0] = ResolveUntilLastSymlink(os.Args[0])\n\n\trun()\n}\n")
+var bbMainSource = []byte("// Copyright 2018 the u-root Authors. All rights reserved\n// Use of this source code is governed by a BSD-style\n// license that can be found in the LICENSE file.\n\n// Package main is the busybox main.go template.\npackage main\n\nimport (\n\t\"errors\"\n\t\"log\"\n\t\"os\"\n\t\"path/filepath\"\n\n\t\"github.com/u-root/gobusybox/src/pkg/bb/bbmain\"\n\t// There MUST NOT be any other dependencies here.\n\t//\n\t// It is preferred to copy minimal code necessary into this file, as\n\t// dependency management for this main file is... hard.\n)\n\n// AbsSymlink returns an absolute path for the link from a file to a target.\nfunc AbsSymlink(originalFile, target string) string {\n\tif !filepath.IsAbs(originalFile) {\n\t\tvar err error\n\t\toriginalFile, err = filepath.Abs(originalFile)\n\t\tif err != nil {\n\t\t\t// This should not happen on Unix systems, or you're\n\t\t\t// already royally screwed.\n\t\t\tlog.Fatalf(\"could not determine absolute path for %v: %v\", originalFile, err)\n\t\t}\n\t}\n\t// Relative symlinks are resolved relative to the original file's\n\t// parent directory.\n\t//\n\t// E.g. /bin/defaultsh -> ../bbin/elvish\n\tif !filepath.IsAbs(target) {\n\t\treturn filepath.Join(filepath.Dir(originalFile), target)\n\t}\n\treturn target\n}\n\n// IsTargetSymlink returns true if a target of a symlink is also a symlink.\nfunc IsTargetSymlink(originalFile, target string) bool {\n\ts, err := os.Lstat(AbsSymlink(originalFile, target))\n\tif err != nil {\n\t\treturn false\n\t}\n\treturn (s.Mode() & os.ModeSymlink) == os.ModeSymlink\n}\n\n// ResolveUntilLastSymlink resolves until the last symlink.\n//\n// This is needed when we have a chain of symlinks and want the last\n// symlink, not the file pointed to (which is why we don't use\n// filepath.EvalSymlinks)\n//\n// I.e.\n//\n// /foo/bar -> ../baz/foo\n// /baz/foo -> bla\n//\n// ResolveUntilLastSymlink(/foo/bar) returns /baz/foo.\nfunc ResolveUntilLastSymlink(p string) string {\n\tfor target, err := os.Readlink(p); err == nil && IsTargetSymlink(p, target); target, err = os.Readlink(p) {\n\t\tp = AbsSymlink(p, target)\n\t}\n\treturn p\n}\n\nfunc run() {\n\tname := filepath.Base(os.Args[0])\n\terr := bbmain.Run(name)\n\tif errors.Is(err, bbmain.ErrNotRegistered) {\n\t\tif len(os.Args) > 1 {\n\t\t\tos.Args = os.Args[1:]\n\t\t\terr = bbmain.Run(filepath.Base(os.Args[0]))\n\t\t}\n\t}\n\tif errors.Is(err, bbmain.ErrNotRegistered) {\n\t\tlog.SetFlags(0)\n\t\tlog.Printf(\"Failed to run command: %v\", err)\n\n\t\tlog.Printf(\"Supported commands are:\")\n\t\tfor _, cmd := range bbmain.ListCmds() {\n\t\t\tlog.Printf(\" - %s\", cmd)\n\t\t}\n\t\tos.Exit(1)\n\t} else if err != nil {\n\t\tlog.SetFlags(0)\n\t\tlog.Fatalf(\"Failed to run command: %v\", err)\n\t}\n}\n\nfunc main() {\n\tos.Args[0] = ResolveUntilLastSymlink(os.Args[0])\n\n\trun()\n}\n\n// A gobusybox has 3 possible ways of invocation:\n//\n// ## Direct\n//\n//   ./bb ls -l\n//\n// For the gobusybox, argv in this case is [\"./bb\", \"ls\", \"-l\"] on all OS.\n//\n//\n// ## Symlink\n//\n//   ln -s bb ls\n//   ./ls\n//\n// For the gobusybox, argv in this case is [\"./ls\"] on all OS.\n//\n//\n// ## Interpreted\n//\n// This way exists because Plan 9 does not have symlinks. Some Linux file\n// system such as VFAT also do not support symlinks.\n//\n//   echo \"#!/bin/bb #!gobb!#\" >> /tmp/ls\n//   /tmp/ls\n//\n// For the gobusybox, argv depends on the OS:\n//\n// Plan 9: [\"ls\", \"#!gobb!#\", \"/tmp/ls\"]\n// Linux/Unix: [\"/bin/bb\", \"#!gobb!#\", \"/tmp/ls\"]\n//\n// Unix and Plan 9 evaluate arguments in a #! file differently, and, further,\n// invoke the arguments in a different way.\n//\n// (1) The absolute path for /bin/bb is required, else Linux will throw an\n//     error as bb is not in the list of allowed interpreters.\n//\n// (2) On Plan 9, the arguments following the interpreter are tokenized (split\n//     on space) and on Linux, they are not. That means we should restrict\n//     ourselves to only ever using one argument in the she-bang line (#!).\n//\n// (3) Which gobusybox tool to use is always in argv[2].\n//\n// (4) Because of the differences in how arguments are presented to #! on\n//     different kernels, there should be a reasonably unique magic value so\n//     that bb can have confidence that it is running as an interpreter, rather\n//     than on the command-line in direct mode.\n//\n// The code needs to change the arguments to look like an exec: [\"/tmp/ls\", ...]\n//\n// In each case, the second arg must be \"#!gobb!#\", which is extremely\n// unlikely to appear in any other context (save testing files, of course).\n//\n// The result is that the kernel, given a path to a #!gobb#! file, will\n// read that file, then exec bin with the argument from argv[2] and any\n// additional arguments from the exec.\nfunc init() {\n\t// Interpreted mode: If this has been run from a #!gobb!# file, it\n\t// will have at least 3 args, and os.Args needs to be reconstructed.\n\tif len(os.Args) > 2 && os.Args[1] == \"#!gobb!#\" {\n\t\tos.Args = os.Args[2:]\n\t}\n\n\tm := func() {\n\t\tif len(os.Args) == 1 {\n\t\t\tlog.Fatalf(\"Invalid busybox command: %q\", os.Args)\n\t\t}\n\t\t// Use argv[1] as the name.\n\t\tos.Args = os.Args[1:]\n\t\trun()\n\t}\n\tbbmain.Register(\"bbdiagnose\", bbmain.Noop, bbmain.ListCmds)\n\tbbmain.RegisterDefault(bbmain.Noop, m)\n}\n")

test/invoke/.gitignore

@@ -0,0 +1,5 @@
+/bb
+/bb-on
+/bb-auto
+/bb2
+/makebb

test/invoke/mod/cmd/helloworld/.gitignore

@@ -0,0 +1 @@
+helloworld

test/invoke/mod/cmd/helloworld/hello.go

@@ -0,0 +1,9 @@
+package main
+
+import (
+	"fmt"
+)
+
+func main() {
+	fmt.Println("hello world")
+}

test/invoke/mod/go.mod

@@ -0,0 +1,3 @@
+module github.com/u-root/gobusybox/test/invoke/mod
+
+go 1.13

test/invoke/test.sh

@@ -0,0 +1,30 @@
+#!/bin/bash
+set -eux
+
+(cd ../../src/cmd/makebb && GO111MODULE=on go build .)
+MAKEBB=../../src/cmd/makebb/makebb
+
+for GO111MODULE in on auto;
+do
+  GO111MODULE=$GO111MODULE $MAKEBB -o bb-$GO111MODULE ./mod/cmd/helloworld
+  test -f ./bb-$GO111MODULE
+
+  HW=$(./bb-$GO111MODULE helloworld);
+  test "$HW" == "hello world" || (echo "hello world not right: direct invocation failed" && exit 1)
+
+  ln -s bb-$GO111MODULE helloworld
+  HW=$(./helloworld)
+  test "$HW" == "hello world" || (echo "hello world not right: symlink invocation failed" && exit 1)
+  unlink helloworld
+
+  # add an interpreter file that contains #!/bin/bb #!gobb#!
+  echo "#!$(pwd)/bb-$GO111MODULE #!gobb!#" > helloworld
+  chmod +x helloworld
+  HW=$(./helloworld)
+  test "$HW" == "hello world" || (echo "hello world not right: interpreter invocation failed" && exit 1)
+  unlink helloworld
+done
+
+# check reproducible
+cmp bb-on bb-auto
+rm bb-on bb-auto